Hi Steven,
you don't mention what OS you are running but in general with 64bit
machines it is good to double the old 32 bit recommendations as a first
pass in server tuning.
For 2008 R2 the OS takes a gig or more with its trick memory free space
caching and its a good starting point with CF to set Max Perm at 512MB
and Max Heap at 2048MB so you need at least 3GB allocated to your VM.
As soon as the sites get busy you will have to up those numbers. If you
have CF Enterprise then turn _off_ the server memory monitoring as it
won't let you go above 2GB Heap and you will want to. It will force a GC
when you don't want one.
Our busy "ordinary" servers run with 5GB RAM, 3GB Max Heap and with the
OS free space caching they run with about 100MB of real free memory
space, and run fast...
With a stored proc running for 300 seconds I would also be looking at
how the DB server is configured, that is a lonnng proc.
HTH
Kym K
On 27/01/2012 10:21 PM, Stephen M wrote:
We recently moved one of our CF projects to Amazon Web Servers and
have started to get the "500 internal server error". This happens
regularly when we run a stored proc with does a lot of DB processing.
It often takes more than 300 seconds. I've set all the CF timeouts to
300 secs so why don't I at least get a CF exception that I can
handle?
This app previously ran on CF7 and it never produced the 500 error.
If it timed out then we got a CF timeout that I could handle with
code. But that was a local pair of VM servers. We also get this
problem on our UAT environment which is local (a pair of VM servers in
our building), so I shouldn't be blaming Amazon.
This is a fairly new CF install. I had originally left the JVM heap
and Garbage collection settings at default. Do they typically require
modifications when installing on 64 bit systems? I've just changed
the UAT server to JVM heap min& max of 1024MB but it hasn't helped.
I have set production at 512 and 768.
There's not a lot of CF processing going on. It just displays the
query result in a table. And we rarely have more than 3 or 4 users at
a time. But these errors are occuring when I know that I am the only
user. So there's one stored proc being called and CF is just sitting
there waiting. Often the error will occur, then I hit refresh and
bang, the query result displays perfectly. Hmmm, that final cfloop to
display the table is pretty big but it hasn't been a problem in the
past.
I have looked at all those JVM heap and GC settings in the past for an
app that does have lots of users and does a lot of CF processing and
it certainly helped but I really don't think that CF is the problem
here. There just aren't that many users and not a lot of
processing.
Anyway, we're running out of ideas, I'll listen to anything to do with
"500 internal server error"
regards,
Stephen
--
Yours,
Kym Kovan
mbcomms.net.au
--
You received this message because you are subscribed to the Google Groups
"cfaussie" group.
To post to this group, send email to cfaussie@googlegroups.com.
To unsubscribe from this group, send email to
cfaussie+unsubscr...@googlegroups.com.
For more options, visit this group at
http://groups.google.com/group/cfaussie?hl=en.
